When a large number of processes need to send information to a small number of processes, or even to a single collector, standard communication procedures call for the establishment of either a connection between all the sending processes to the collector, or to at least open a unicast address between each of the senders and the collector. This typically causes an unnecessarily complex communications scheme on both the sender and the receiver sides. The number of connections, or unicast channels, is N (number of senders)×M (number of receivers).
Creating a high availability model for such an M×N communication model is complex, and typically requires significant monitoring, since the unicast addresses are not anonymous and any change in the location of the receiving process must be known to the sending processes, preferably in advance.
There is thus a need for addressing these and/or other issues associated with the prior art.